Introduction

Helping the children of new settlers adapt to their schools, integrate with the larger society and stay happy and healthy during the process are important goals for all immigrant receiving countries. However, the current dearth of knowledge about what promotes adaptation and integration and what jeopardizes the well-being of immigrant and refugee children hampers the development of effective policy and practice [11, 16, 51].

To help rectify the situation, information is required about the extent to which factors that seem to affect the mental health of most, if not all children, are also salient for children in immigrant families. There is also a need to identify risk and protective factors unique to the experience of children in immigrant families. The New Canadian Children and Youth Study (NCCYS) is a longitudinal investigation of personal and contextual factors affecting immigrant children’s mental health. The current report, based on data from the study’s initial survey, has four major objectives: (1) explore the extent to which determinants of mental health for children in general are also relevant for immigrant children, (2) identify which, if any immigration and settlement experiences affect the mental health of immigrant children, (3) explore the mental health salience of place of resettlement, (4) demonstrate the importance of unpacking overly inclusive categories such as “Asian immigrants.”

Background

One in five children living in Canada was born either outside the country or to recently arrived immigrants [16, 48]. Insights gleaned from studying children in the general population (see for example [49]) probably tell only part of the story about immigrant children and their mental health. Migration and resettlement create unique challenges that call for the development of special personal and social resources to meet them.

Mental health is the product of interactions among personal factors such as age, gender, and ethnicity, and ecological influences, such as household characteristics, income, parental mental health, family functioning, and neighborhood quality [15]. Since the mental health salience of a risk factor such as poverty and a protective factor such as good family functioning has been repeatedly confirmed, it is hard to imagine omitting them from any serious inquiry into children’s well-being. In acknowledgement of this history, the current report alludes to these factors as mental health “universals.” However, since one can never be sure that anything is truly universal, we add the adjective “putative.” Over and above the challenges all children face, immigrant children have to deal with relatively unique stressors, and they may benefit from relatively specific protective factors [7, 16, 27, 32, 40]. The current report explores the mental health salience of both putatively universal and immigration-specific processes.

Bronfenbrenner’s [15] concept of nested ecological systems provides a theoretical framework within which to identify factors ranging from the micro- to the macrosystem level that can affect mental health. Parental depression, a proximal, microsystem variable, jeopardizes the mental health of both native-born [22, 29, 41, 54, 55] and immigrant children [40]. Among children in the general population, poverty has a directly deleterious effect on mental health, as well as indirect effects through linkages with ineffective parenting, parental psychopathology, intrafamilial hostility, single parent families, and overcrowded housing [55, 56]. According to Canadian studies [7, 27], poverty affects immigrant children, just as it does their majority culture counterparts. However, the effect is smaller, probably because poverty among immigrants is less likely to be linked to single parent families or intrafamilial hostility [7].

More than one-third of adult immigrants speak neither English nor French on arrival in Canada [48]. Research [58] suggests that parental inability to speak the language of the receiving society jeopardizes children’s mental health. One possible reason is that lack of linguistic fluency impedes parent’s ability to take part in their children’s socialization, most particularly in their schools [58]. Another problem is that children’s ability to learn a new language more quickly than their parents can create role reversal, a situation in which children are called upon to negotiate with the larger society on their parent’s behalf, thereby being forced into assuming roles usually reserved for adults [16, 32, 39]. More distal factors, such as dysfunctional neighbourhoods have been implicated as mental health risk factors for all children [23, 28]. Although not exclusive to the immigrant experience, prejudice is a mental health challenge confronted by many immigrant children. One in five children of new settlers—particularly those who belong to visible minority groups—reports encounters with discrimination and prejudice during their resettlement in Canada [4, 9, 10, 42]. On the protective side of the ledger, the like-ethnic community, so important to maintaining the well-being of adult refugees in resettlement countries [9] is a possible mental health resource for their children as well.

Countries, and even regions within a country, offer newcomers differing “levels of hospitality,” that is macrosocial climates that can affect mental health [3, 9, 16]. Inter-provincial disparities in the amounts of money spent per immigrant [16, 20] translate into differential access to language training, day care, job training programs and health care, each of which may affect the well-being of parents and children. Investigations focusing on newcomer children often employ overly broad categories such as “Asian immigrant [27].” One of the aims of the current study is to demonstrate the importance of unpacking such categories. To do so, country of origin is used as an admittedly imperfect proxy for ethnocultural background. This is not meant to imply that people from a particular country of origin are ethnically homogeneous, but rather that people coming from different countries have been exposed to relatively unique cultural influences during their pre-migration lives, that they follow different routes to get to Canada, and that country of origin affects the reception given to them on arrival.

The present report is concerned with children and their families from three different immigrant groups—Hong Kong (HK) Chinese, Mainland (PRC) Chinese and Filipino. Although immigrants have been coming from Hong Kong to Canada since the late nineteenth century, the flow increased dramatically during the 1980s and early 1990s, in anticipation of the 1997 UK handover of Hong Kong to China. Although most Hong Kong Chinese families entered Canada under the country’s “point system,” under which immigration applicants can be admitted if they achieve a certain score based on factors such as education, English and French language fluency, occupational skills, and presence of family members in Canada, some applied for admission under Canada’s entrepreneurial class, a special category designed to attract people with financial skills and resources [20]. Although they acquired Canadian citizenship, many families chose to retain HK citizenship in order to pursue business opportunities at home. The so-called “astronaut” or transnational family, in which parents spend long periods of time in Hong Kong while the children stay in Canada in the care of domestic help, has been a product of these opportunities and choices [35, 44, 45, 52]. International emigration from Mainland China began in the early 1980s, after the relaxation of tensions between the Peoples Republic of China (PRC) and the west. In comparison with their Cantonese-speaking counterparts from Hong Kong, PRC Chinese immigrants are more likely to speak Mandarin, have fewer economic resources, and live in more residentially stable families [19, 20, 36, 37]. Both HK and PRC Chinese live in Canada’s largest cities, where they remain the most residentially segregated of all of Canada’s visible minority groups [48]. Most (60%) of Philippines-origin immigrants are women, and most—because of the century-long US presence in their country—were educated in English [17, 18]. Canada’s shortage of nurses and of trained personnel to care for children and the elderly created a portal of entry that attracted many Filipina immigrants. The country’s Live-in-Caregiver program provides fast tracking to permanent resident status in return for an initial 24 months of paid service as a caregiver for children or the elderly. The Filipino pattern of immigration gives rise to transnational families very different from HK “astronauts.” Filipino immigrants often come alone, share wages with family left behind, and then, after achieving permanent resident status, sponsor family members to join them [1, 2, 17, 18, 34].

Despite inter-group differences, Asian immigrants are highly likely to experience prejudice as a common problem. As many as 25% of Asian immigrants report encounters with prejudice within the first 10 years of resettling in Canada [4, 9, 10, 42]. Aside from visible minority status, the social construction of immigrant is freighted with negative attributions [9].

Methods

The NCCYS is a longitudinal study, with two age-defined inception cohorts: children 4–6, and children 11–13. Children in the sample were either born abroad or into families in which at least one parent had immigrated to Canada within a 10-year period prior to the study’s inception. Since almost all immigrants settle in cities [48], samples were recruited from the major cities in four federally defined regions: Vancouver, in the Pacific region; Edmonton, Calgary and Winnipeg, Western region; Toronto, Central region; and Montreal, Eastern region.

Study sample

A sub-study of the NCCYS, the current investigation aimed to recruit a sample of 180 children in each region, stratified to provide equal numbers by ethnicity, and age-group (4–6 and 11–13), for a total of 2,160. A number of strategies recommended for difficult to identify, hard to find, and highly mobile groups [50, 53] were employed. In the six cities, representatives from each of the targeted ethno-cultural groups were invited to become members of Community Advisory Councils (CAC). After privately identifying families from their respective communities who fit NCCYS inclusion criteria, CAC members contacted potential sample families to request permission to introduce a research team member to explain the study and to obtain written consent to participate. Each participant was asked to think of other families with children meeting the study’s inclusion criteria. They were then asked to contact these families to request permission for contact by the research team.

As Table 1 shows, the sample fell slightly short of its recruitment goals, particularly in its representation of the younger and the Hong Kong Chinese cohorts. Based on a population model using 2001 census data, each sample was weighted to the Census Subdivision (CSD) level. The CSD is usually a provincially defined municipality.

Table 1 Sample description
Full size table

Measures

The NCCYS questionnaire elicits information about parental ethnic and religious backgrounds, parents’ education, labor force activity, income, health status, family functioning, parenting behaviors, social support, neighborhood characteristics, pre- and post-migration stressors, relationships with the like-ethnic community, relationships with larger societal institutions, the child’s general health and developmental history, and indicators of emotional problems (EP) and physical aggression (PA).

Canada’s National Longitudinal Study of Children and Youth (NLSCY) [49] provided a framework for the development of the NCCYS measures of mental health, as well as of many of the putatively universal risk and protective factors. CAC’s in each of the study regions judged the acceptability, the likelihood of achieving appropriate translation, and face validity of each of the NLSCY items. The immigrant-specific portions of the NCCYS derive either from the immigrant mental health literature, or were developed by the NCCYS team. NLSCY-derived items that survived the preliminary selection process, together with the immigrant-specific items were assembled into a draft questionnaire, translated into each of the heritage languages, and then back-translated in order to uncover and resolve ambiguities. Before embarking on the survey, bilingual interviewers, many of whom were immigrant or refugee professionals, took part in a 2-day training session. To ensure standardization, one senior member of the investigative team assumed responsibility for training in all of the centres. Project personnel interviewed the person most knowledgeable (PMK) about the child. In most cases, this was the child’s mother. Pilot testing with the draft questionnaires provided another opportunity to identify troublesome items and to resolve ambiguities. Items that survived the two-stage process of community consultation and pilot testing were retained for the study.

The dependent variables for this report are EP and PA. Predictors include sociodemographic variables, putatively universal mental health risk and protective factors, and migration-specific risk and protective factors. Reliability coefficients (Cronbach’s alpha) were calculated for each scale. In order to help overcome the problem of missing data, response choices for the individual items making up a scale were averaged to form a continuous measure.

Dependent variables

(a) Emotional problems (EP), measured by an eight item scale derived from the Ontario Child Health Survey [5, 13, 14] and the NLSCY [49]. Each item has three forced-choice responses: (1) (never or not true), (2) (sometimes or sometimes true), and (3) (often or very true). Sample items include: How often would you say that (index child) (i) seems unhappy, sad or depressed? (ii) not as happy as other children? Reliability coefficient 0.75. Due to the use of item means rather than raw scores, scale scores range from 1 to 3, rather than 3 to 24. Mean 1.310, SD 0.296. (b) Physical aggression (PA), based on the same sources as EP [5, 13, 14, 49], contains six items, with stem question and forced-choice responses identical to the EP. Sample items include: (i) gets into many fights? and (ii) is cruel, bullies or is mean to others? Reliability coefficient 0.74, range 1–3, mean 1.165, SD 0.255.

Putatively universal sociodemographic influences on mental health

Variables include (a) age cohort (measured by a dummy variable, ref. category = 11–13); (b) gender, (a dummy variable, ref. category = male); (c) parental education, PMK with university degree = 1, lower than university degree = 0; (d) parent’s marital status: married = 1, other = 0.

Putatively universal risk and protective factors

Putatively universal mental health risk factors include: (a) low income—a measure of household income adequacy developed, and routinely used by Statistics Canada. Low income is defined as a household with 1–2 persons and a total income below $15,000, or a 3–4 person household with income below $20,000, or a household with 5 or more persons and income below $30,000. When household income was not reported, we used a dummy variable representing income missing. For both low income and missing income, the common reference groups is families above the low-income thresholds. This permitted us to retain cases with missing data, while avoiding unwarranted assumptions necessary for imputing values [57]. (b) PMK depression, based on forced-choice responses (rarely, sometimes, occasionally or most of the time) to 16 questions developed for use among Southeast Asian [6, 9]: the stem question is “Have you…” Item examples include: (i) been feeling unhappy? (ii) been feeling discouraged? (iii) low and hopeless? Reliability coefficient 0.90, range 1–3.75, mean 1.546, SD 0.444.

Putatively universal mental health protective factors included: (a) positive family functioning: fixed response categories for this 12 item scale [27, 49] range from 1 (strongly disagree) to 5 (strongly agree). Item examples include: (i) we express feelings to each other (ii) all members of the family are accepted regardless of their character. Reliability 0.85, range 1–5, mean 3.938, SD 0.392. (b) Neighbourhood cohesiveness: derived from the NLSCY [49], the scale contains five items, each with a response choice of 1 through 5 (strongly disagree to strongly agree). Item examples include: (i) if there is a problem around here, the neighbours get together to deal with it (ii) people around here are willing to help their neighbours. Reliability coefficient 0.81, range 1–5, mean 3.422, SD 0.540, and (c) PMK perceived social support: This seven item scale [49], has five forced-choice responses ranging from strongly agree to strongly disagree. Item examples include: (i) I have family and friends who help me feel safe, secure and happy (ii) there is someone I trust whom I would turn to for advice if I were having problems. Reliability coefficient 0.88, range 1–5, mean 3.579, SD 0.588.

Migration-specific sociodemographic, risk, and protective factors

Migration-specific sociodemographic variables include: (a) country of origin, measured by dummy variables, (reference category = Filipino); and (b) region of resettlement measured by dummy variables, (reference category = Toronto). Risk factors specific to the immigrant experience include: (a) PMK’s ability to speak English or French fluently: responses for the sample as a whole were (Yes 24%; no 76%) (b) PMK resettlement stress: adapted from the Southeast Asian Refugee Resettlement Project [9] and further developed by working with community advisory groups, this scale contains seven items with responses ranging from never = 1 to very often = 4. The stem question is “How often have you felt stress because of….” Item examples: (i) I am living away from my family, relatives and friends (ii) I am unable to do the things I used to enjoy when I was in…. (country of origin). Reliability coefficient 0.90, range 1–4, mean 1.760, SD 0.437. (c) PMK perception of prejudice. A six item scale, derived from the Refugee Resettlement Project [10, 42], with response choices from 1 through 5 (strongly disagree to strongly agree). Item examples include: (i) A lot of Canadians tend to look down on people from my country (ii) Landlords in Canada would rather rent an apartment to someone from another group than someone from my country. Reliability coefficient 0.80, range 1–6, mean 2.655, SD 0.583. The putative protective factor, like-ethnic neighbourhood concentration, was measured by one item, “This neighbourhood has many people from my country of origin,” with five responses ranging from strongly agree to strongly disagree. The item was recorded as follows: strongly agree and agree = Yes (60%); all other responses = No (40%).

Results

Results described in Table 2 suggest differences in levels of emotional disorder and PA among children in the three ethnocultural groups, that PMK’s from Hong Kong were less well-educated than their counterparts from the PRC and the Philippines, and that Filipino PMK’s were more likely than the PMK’s in the other two groups to be fluent in English or French. PRC and HK Chinese lived in more ethnically dense neighbourhoods than Filipinos. Compared to the two other groups, PRC households were more likely to be poor.

Table 2 Study variables by ethnic group
Full size table

Predictive models were tested for each of the two dependent variables. The EP and PA scales, respectively, were regressed on demographic and on putatively universal risk and protective factors on step 1. Step 2 added immigration and resettlement-specific predictors including ethnicity and place of resettlement.

Table 3 examines results for EP. Model 1 accounted for 12.1% of the variance in scores. The addition of immigration-specific variables in Model 2 increased predictive power by 3.5%. PMK depression and familial dysfunction made significant and independent contributions to EP. According to Model 2, HK children were at higher risk for EP than Filipino or PRC youngsters. PMK inability to speak one of Canada’s official languages, resettlement stress, and perceptions of prejudice each increased the risk for children’s EP. Region also made a difference. Children in Toronto were at higher risk for EP than children in Vancouver, Montreal or the Prairies.

Table 3 Regression of emotional problems on sociodemographic, universal and migration-specific predictors
Full size table

Table 4 presents the results of regressing PA on the predictor variables. Model 1 accounts for 6.3% of the variance in risk of PA. Adding the immigrant-specific stress and protective variables in Model 2 improved the predictive power by almost 3%. In contrast with EP, where child characteristics had no apparent effect, younger children were more likely than older to be physically aggressive, and boys more likely to display PA than girls. PMK depression was, once again, a significant predictor. Compared with Filipino children, PRC Chinese displayed less PA, and HK Chinese, more. Children in Toronto were at higher risk for PA than children in Vancouver. Although differences for the other two regions did not achieve statistical significance, the trend was similar to that observed for EP, with Toronto the setting characterized by the highest level of risk. Acculturation stress and perceived prejudice made significant and independent contributions to explaining physical aggressiveness scores.

Table 4 Regression of physical aggression on sociodemographic, universal and migration-specific predictors
Full size table

Discussion

Methodological limitations must be acknowledged. The findings are based on three groups selected because, at the time the NCCYS began, they made up the largest proportion of immigrant children in Canada. They do not constitute a representative sample of all immigrant children, or even of the three countries of origin groups taking part in this study. The results should not, and cannot be generalized to all immigrant children. True probabilistic sampling of these difficult to locate, highly mobile populations will probably always be elusive. The aim of the NCCYS is not, however, to establish rates and to make cross-sample comparisons, but rather to investigate the impact of theoretically meaningful factors on the mental health of immigrant children through statistical modeling of putative risk and protective factors. Weighting has been used to help compensate for problems such as restricted variability.

There are also limitations stemming from the data and the analyses. The relatively large number of variables included, together with large sample sizes gives rise to the problem of alpha inflation. Rather than adopt a strategy such as restricting the definition of statistical significance to an arbitrary cut-off, this report presents findings using conventional levels of significance, leaving it to the reader to judge the substantive significance of the findings. Although the amount of variance explained compares favorably with amounts reported in other studies of similar populations [23], 15.6 and 9.1% leave a great deal of variance unaccounted for. Future NCCYS publications will include a much larger number of immigrant communities, thereby permitting examination of potentially important mental health risk factors such as separation from parents [11], and immigrant versus refugee status [16]. The fact that the PMK was the sole source of information may be another limitation, since other studies [31, 51] have demonstrated that children’s self-reports evidence limited overlap with parent ratings. Relying on one source of information may result in biased associations. For example, the association between PMK perceptions of prejudice and childhood distress may be a substantive finding reflecting the adverse effect of prejudice on everyone in the family, but it is also possible that the relationship is a confound stemming from the PMK’s projection onto children of their own distress and their attributions regarding cause [51]. Future NCCYS reports will include children’s self-reports, as well as data from their parents. Finally, because data for the current report are cross-sectional, they cannot address the question of causal sequencing. It is, for example, plausible that physically aggressive children disrupt family functioning, rather than the other way round. Future analyses employing NCCYS longitudinal data can help address this problem.

It seems that, in many ways, immigrant children’s mental health is affected by the same factors that affect the mental health of children in general. For example, consistent with results from other studies, boys are more likely than girls, and younger children more likely than older, to display PA [22, 33, 41, 43]. As is the case for children in general, parental depression increases the risk for childhood EP [24, 29]. However, in contrast with studies that have focused on “mainstream” populations [23, 28], neighbourhood characteristics had no demonstrable effect on the mental health of immigrant children. Perhaps the neighbourhoods in which immigrants live during the early years of resettlement do not offer sufficient variability to demonstrate statistically significant associations. A more substantive explanation is also possible. Some authorities [25, 27] have suggested that family factors trump neighbourhood characteristics as predictors of children’s mental health.

Consistent with other reports [7, 57], this study demonstrates that, despite the stresses of resettlement, immigrant children manage to stay in relatively good mental health. Low income, one of the most powerful mental health risk factors for children in general [7, 54, 55], had no significant association with either EP or PA among HK, PRC and Filipino immigrant children in this study. Previous research reports [7, 27] have noted that poverty had a less powerful effect on children’s mental health among immigrant, than among native-born children. One possible interpretation is that poverty is a different experience for immigrants and native-born children and their families. Single parent and dysfunctional family status, for example, are frequent concomitants of poverty among native-born, but not among immigrant families [7].

The PMK’s in these samples had remarkably low levels of proficiency in English or French. According to official reports [20], about one-third or more of immigrants speaks neither language on arrival. The rates for PMK’s in this study were much higher, ranging from 50 to almost 80%. One reason for the discrepancy is that most of the PMK’s were mothers. In most immigrant families, the father is the principal applicant, and he must achieve a certain score in order to qualify for admission. Language proficiency is an important component of this score. Other members of the immediate family do not have to pass the so-called “points” test. Lack of linguistic fluency among mothers apparently had an adverse effect on children, perhaps because lack of language compromises adult mental health [16] or perhaps because it gives rise to role reversal—a situation in which children, who learn the language and culture of resettlement countries more quickly than their parents, may be called upon to negotiate with the larger society in ways usually reserved for adults [16, 47]. Canada provides English or French language training for newcomers, but the limited duration of the classes, inappropriate mix of students, and barriers limiting women’s participation [8, 13, 21] compromise program effectiveness.

In view of the many studies attesting to the importance of the like-ethnic community in supporting the mental health of adult immigrants [3, 9], the lack of association between this variable and children’s mental health is surprising. Perhaps the operational measure of ethnic concentration failed to do justice to the underlying construct. Another possibility is that “neighbourhood” in this study referred to geographical community, whereas people may define community in other ways that are more relevant for their mental health. Finally, a like-ethnic presence may be less salient for young children than it is for adolescents and adults.

One of the study’s aims was to investigate possible mental health effects of region of resettlement. Given Toronto’s reputation as a multicultural city, it is surprising that it is the place in which immigrant children apparently experience the greatest mental health risk. Previous research [12] demonstrating that, in comparison with other provinces, Quebec was less accepting of immigrants would lead to the prediction that Montreal would occupy the position Toronto assumes in the current study. However, that study’s data are now more than 15 years old, and secular changes may have affected attitudes towards immigrants. In the early 1990s, immigration began acquiring increasing cachet in Quebec, the Prairies and British Columbia. For example, in 1991, the federal government signed an accord with Quebec, devolving jurisdiction as well as funding for settlement and integration to the Province. Similar accords were signed with Manitoba in 1996, with British Columbia in 1998, and with Alberta in 2002. By contrast, during the mid- to late 1990s, Ontario provided severely limited amounts of the kinds of social support that many immigrant families require during the early years, remained cool towards immigrants, and suspicious of federal policies of devolution. It was not until 2004 that the province signed an initial letter of intent to proceed with negotiations regarding immigrant selection, destination and integration. Despite being the largest magnet for immigrants, Ontario may not have presented the most welcoming environment. Future NCCYS studies will attempt more in-depth explorations of the relationships between region of resettlement and children’s well-being.

The study’s results demonstrate the importance of unpacking overly broad categories such as “immigrant children” or even “Asian immigrants.” PRC children experienced a lower risk of developing problems than either HK or Filipino youngsters. These findings call attention to the circumstances of migration, in particular the phenomenon of transnational families. Filipino migration is often initiated by women who travel abroad to respond to the needs for trained personnel by some countries or to take advantage of special situations such as Canada’s “fast track” path to landed immigrant status. When family reunification takes place, often after a prolonged separation, it can be complicated by children’s resentment over perceived maternal abandonment [1, 11, 45]. Many HK families apparently come to Canada with plans to stay long enough to ensure their children’ education, but with the ultimate goal of returning to the home country [35, 44]. Authorities [11, 44, 52] have raised concerns about the possible mental health consequences of prolonged parental absences, and of pursuing the goal of return rather than of integration. Although it is tempting to speculate that the increased mental health risk among HK Chinese and Filipino children may be at least partially attributable to parental absences consequent on transnationalism, drawing such conclusions would be premature. For one thing, such an interpretation would be open to the ecological fallacy—to attributions based on group characteristics that may or may not apply to study subjects themselves. Future NCCYS studies will focus on relationships between parental absences and children’s mental health risk. Since family separation is, to a certain extent, amenable to changes in policy, these results cast a potentially important light on the importance of speeding up family reunification. With respect to services, if children in transnational families are indeed subject to particular mental health risk, meeting their needs may call for special training programs for service providers, including the need to plan for family life post-reunification [11].

The over-riding significance of the study lies in demonstrating that migration-specific variables influence the well-being of the children of immigrants, net of factors that affect the mental health of most, if not all children, Future research and intervention programs should take account of the specificity of mental health risk and protective factors on the lives of children in immigrant families